I. Foundations & The Laws of Physics
Distributed systems are not a scaling technique; they are a different physics. The moment state and computation cross a network boundary, guarantees that a single machine gave you for free — a global clock, reliable memory, atomic visibility of writes, all-or-nothing failure — disappear and must be rebuilt from weaker primitives. This part establishes the hard constraints every later chapter inherits: the economics that push you off a single node, the failure models you must design against, and the impossibility of a shared notion of “now.”
Topics Covered
Section titled “Topics Covered”- 1.1. From Monoliths to Distributed Systems: The architectural shift from single-node monoliths to distributed systems, covering scaling strategies, cost curves, and network realities.
- 1.2. System Models and Failure Types: How network timing and failure assumptions form the system model that every correctness argument depends on.
- 1.3. Time and Ordering: Replacing wall-clock intuition with causality: happened-before, logical and vector clocks, hybrid clocks, and TrueTime.